High Specific Energy Lithium Primary Batteries as Power Sources for Deep Space Exploration

Exploration missions to the moons of the outer planets (such as Europa) pose unique challenges regarding the design of the spacecraft power source. Current aerospace qualified primary battery technologies cannot adequately meet the mass and volume requirements of proposed missions. Although they have not been used in prior deep space landed missions, lithium carbon-fluoride (Li/CFx) technologies were identified as a potentially viable option, both with and without blends of manganese dioxide (MnO2). To meet the performance requirements over the intended operating conditions of future NASA missions requires further development of this technology, in particular in the delivery of a high specific energy at moderate to low temperatures, and low discharge rates. A cell development effort was therefore pursued with an industrial battery cell manufacturer. Low (50 mA) and medium (250 mA) discharge rates were used to assess the performance of D-size cells under mission relevant conditions, between 0◦C and −40◦C. Select AA-size and C-size cells were also evaluated using similar rates scaled to the lower cell capacities. Developmental Li/CFx-MnO2 D-size cells designed for higher specific energy over these conditions were fabricated and tested, targeting operation between 0 and −40◦C and a 50 mA constant discharge current, as the baseline operating condition. © The Author(s) 2018. Published by ECS. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any medium, provided the original work is properly cited. [DOI: 10.1149/2.1061810jes]

There is increasing interest in the exploration of the moons of the outer planets known as "Ocean Worlds," where significant quantities of subsurface liquid water exist beyond Earth. 1 Of particular interest are Jupiter's moon Europa and Saturn's moons Enceladus and Titan. Recent mission concept studies performed by the National Aeronautics and Space Administration (NASA) and the European Space Agency (ESA) have focused on landed missions to Europa. There is a significant motivation to conduct in situ science on this moon, in particular sampling of the surface and sub-surface ice with subsequent analysis for habitability, bio-signatures and even extant life. Supporting the desired sampling and analysis activities would require significant levels of energy.
Due to Europa's significant distance from the Sun (∼780 × 10 6 km), photovoltaic-based power generation is not feasible for current surface mission concepts. In addition, radioisotope power systems are difficult to implement in the currently planned mission architecture, due to constraints imposed by the baseline lander design and the high radiation environments prevalent on Europa. Given these limitations, the only viable power source of sufficient maturity is a primary battery. Primary batteries have been used in previous in situ exploration missions to Jupiter in 1995 (Galileo Probe), 2 Mars in 1999 (Deep Space 2 Microprobes) 3 and Saturn in 2005 (Huygens Probe). 4 However, unlike these past missions which were required to operate for <24 hours, future missions (such as a proposed Europa Lander) could require surface operations to be conducted for several weeks (∼480 hours). Different lander mission architectures are under investigation, some requiring cold operations (∼−40 • C) and others operating closer to Earth ambient (+20 • C) conditions. Common spacecraft designs place the batteries in thermal contact with the cruise stage propellant loop, enabling battery storage and initial operating temperatures of approximately 0 • C.
Recent projections of the total energy required to perform an adequate science mission are in the range of 35-45 kWh. Peak power demands are anticipated at 500-600 W (during telecommunications events), with quiescent periods requiring 30 W or lower, and average power levels of approximately 60 W. These emerging requirements highlight the fact that many mission concepts would be constrained by available energy, and not power levels. A high specific energy (>700 Wh/kg) primary power source is needed, due to the high sensitivity of lander designs (500-600 kg total) to added mass.
New high specific energy primary battery cell designs based on the Li/CF x -MnO 2 chemistry have recently been reported, specifically designed for improved low temperature performance. 5 Efforts were initiated to select and benchmark existing and emerging primary battery chemistries from several different vendors, and to develop energy storage options needed to fill identified technology gaps. As mission success hinges on the ability to deliver adequate energy and power to support the key science objectives, significant efforts were expended in identifying appropriate technology options. Current tentative plans propose the launch of a landed mission to Europa in the mid-2020s timeframe, followed by a landing in the early 2030s. Initial battery cell options were identified, based on vendor-reported values and their potential to support the emerging architectures for this planned mission. A summary of their reported capabilities are listed in Table I.
The initial assessment of available primary battery cell technology options focused on selection of both liquid cathode (Li/SO 2 and Li/SOCl 2 ) and solid cathode (Li/MnO 2 , Li/CF x , Li/CF x -MnO 2 , and Li/FeS 2 ) systems. Specific cells evaluated as part of this study are listed in Table I. The critical temperature range of interest was identified as 0 to −40 • C. Although several cell formats were investigated based on available chemistries from various vendors, D-size cells were found to offer the best combination of commercial availability, high specific energy and optimal cell geometry to support battery pack designs. Using notional pack designs featuring the D-size cell format, coupled with assumed peak power levels to estimate maximum current levels per cell, a baseline operating condition at the cell level of 0 • C and 50 mA was identified. Testing was ultimately performed on D-size cells between 0 and −40 • C and 50 mA to 250 mA, to encompass a comprehensive range of potential operating conditions.

Experimental
A range of available cells representing a cross-section of chemistries (Table I) were procured from vendors. Wires were soldered to the terminals or tabs (whichever was provided by the vendor), and each cell was placed into a holding fixture. A J-type thermocouple was affixed to the surface of each cell using Kapton tape. Electrical leads were interfaced to a Maccor 4000 automated battery test unit (5V and 7.5 A maximum current per channel). Each cell was placed in a thermally controlled chamber (ESPEC Platinous EPZ-2H) and the chamber brought to the set point temperature. Prior to constant current discharge, each cell was held at the test temperature for a minimum of eight hours to establish thermal equilibrium. Voltage and current data were collected during discharge, along with the cell skin temperature. The test temperature was actively controlled via a thermocouple located near the center of the environmental chamber. D-size cells were evaluated at a low current (50 mA) and a medium current (250 mA) discharge condition, between 0 • C and −40 • C. Li/CF x cells in a Csize format and Li/FeS 2 in an AA-size cell format were discharged at lower absolute current values (low current of 20 mA and medium current of 100 mA), scaled down appropriately from rates used in the D-size format cells. Typical voltage cut-offs were 1.5V, or 0.8V in the case of Li/FeS 2 AA-size cells (which features a nominal discharge voltage of approximately 1.5V, versus 2.5 to 3.0V for the other lithium primary chemistries). The specific energy was determined from the discharge voltage and the delivered capacity, typically averaged from three cells. Impedance was measured at 1 kHz using a Hioki BT-3554 battery tester. Electrochemical impedance spectroscopy (EIS) data were collected over the range of 1 MHz to 125 mHz with 5 mV excitation voltage using a Princeton Applied Research VersaSTAT potentiostat/frequency response analyzer at the DC potential corresponding to the open circuit voltage. The cells were maintained at 21 • C for at least 1 hour prior to EIS analysis using a Tenney TUJR environmental chamber. Developmental Li/CF x -MnO 2 cells were assembled at the vendor facility (EaglePicher). Manganese dioxide (MnO 2 ) was added to the cathode formulation to address some of the historical limitations of carbon fluoride (CF x )-based cells, in particular the observed voltage delay at low temperatures, the high heat generation at higher rates, the safety concerns and the materials cost. Commercially available CF x (x = 1) and electrolytic MnO 2 were used. The MnO 2 was heat treated at temperatures between 350 and 400 • C prior to use, in order to generate the desired polymorph. The CF x -MnO 2 hybrid electrodes were prepared using a web coating process similar to that used for Li-ion battery electrodes. The web coating process enables the production of very thin electrodes for improved rate capability and low temperature performance. The two active materials were mixed with carbon black, graphite and poly(vinylidene difluoride) binder to form a slurry that was then coated on an aluminum current collector. The cathode formulation was modified to achieve a high electrode density with enhanced rate capability. This included optimization of the particle size distribution of the active materials and the porosity of the electrode. The baseline particle size distribution was centered at approximately 38 μm, with the optimized particle size distribution centered at approximately 25 μm. The baseline cell available from the vendor featured a LiClO 4 salt in a solvent blend of propylene carbon-ate, 1,2-dimethoxyethane and tetrahydrofuran (PC:DME:THF), 1:1:1 by volume. For the first cell modification (2R), a reduced LiClO 4 salt concentration was used in the electrolyte formulation, relative to the baseline design. The second cell design modification (3R) combined this reduced salt concentration with a cathode that was thinner relative to the baseline design. The third design variant (4R) featured both of these cell modifications, along with a modified electrolyte solvent formulation using a mono-fluoroethylenecarbonate additive (FEC) in a blend of PC:DME:THF:FEC (15:40:40:5 by volume). The cell electrode assembly was prepared using the same automated winder used for the baseline cells. Aluminum hardware (can and header) were used in place of the traditional stainless steel can, to increase the celllevel specific energy. The use of aluminum hardware increased the specific energy by ∼100 Wh/kg compared to cells using mild steel hardware. Laser welding was used to achieve optimal can-to-header welds for the aluminum cells. Following manufacture, the cells were pre-discharged and conditioned at 55 • C. A final cell screening protocol (involving evaluation of cell mass, open circuit voltage, 1 kHz impedance, and a visual inspection) was performed after a two-week room temperature storage period.

Results and Discussion
Li/SO 2 cells are typically chosen for applications requiring high power delivery at low temperatures, as the charge transfer kinetics for SO 2 dissolved in organic solvent are favorable for low temperature operation; Li/SOCl 2 cells offer higher specific energy but their temperature range and power capabilities are more limited due to the properties of the pure inorganic catholyte used. In contrast, the Li/CF x chemistry is selected where a very high specific energy is required, especially at low rates and moderate temperatures (∼20 • C).The kinetic limitations of the Li/CF x system relate largely to the nature of the solid cathode system (involving diffusion of the Li + ion into the cathode, coupled with the breaking of the C-F bond and the passivation effects from the LiF formation). A number of previous attempts have focused on improving low temperature performance of this chemistry in laboratory cells. [6][7][8] A summary of the specific energy determined for all cell types tested is given in Table II. The mean cell voltages during discharge under various conditions are also given in Figs. 1 and 2; voltage versus discharge capacity curves are shown in Figs. 3 and 4. The temperatures used for discharge, 0 • C and −40 • C, encompass the range of anticipated low temperature operating conditions for the lander application. At 0 • C, the highest specific energy observed was from the Rayovac Li/CF x cells, at both the low current (640 Wh/kg) and medium current (508 Wh/kg) conditions. At −40 • C,   the Li/FeS 2 cells provided the highest specific energy at the low current (276 Wh/kg) and medium current (210 Wh/kg) conditions. At this temperature, the Rayovac Li/CF x cell chemistry was no longer operational and was incapable of supporting these discharge currents, although they were not specifically optimized for low temperature operation. The Panasonic Li/CF x C cells, which likely do not employ sufficiently thin electrodes, provided a much lower specific energy relative to the Rayovac Li/CF x D cells at equivalent currents, and temperatures.
As indicated, Li/CF x cells typically cannot support modest to high currents at reduced temperatures, and often display a significant voltage delay during discharge. To overcome these limitations, several battery vendors have explored "hybrid" designs featuring a composite cathode of CF x and MnO 2 electrode materials. 5,6 These cells display a characteristic two-plateau voltage profile during discharge, with the MnO 2 discharge occurring first at a slightly higher voltage, followed by a lower voltage CF x discharge plateau. Given the lower specific capacity of MnO 2 vs. CF x , this cell chemistry cannot match the higher specific energy of a pure Li/CF x cell. However, the composite electrode can impart the higher current/lower temperature capabilities of the Li/MnO 2 chemistry combined with a still very high specific energy enabled by the pure Li/CF x chemistry. Across the temperature range of 0 • C and −40 • C, the EaglePicher Li/CF x -MnO 2 hybrid cathode cells displayed the highest specific energy, providing 437 Wh/kg and 418 Wh/kg at the low and medium current conditions, respectively, at 0 • C. This cell chemistry also supported discharging down to −40 • C, where specific energies of 226 Wh/kg (low current) and 131 Wh/kg (medium current) were measured.
As indicated above, there was particular interest in improving the specific energy at the low current 0 • C condition, since the cells will likely need to support discharge under these conditions when lander surface operations are initiated. Following operation under this condition for some period, it is anticipated that the battery temperature would increase due to self-heating, bringing the cells into a warmer temperature regime where kinetic limitations are less of a concern. It is possible that the battery temperature will again approach the initial temperature (0 • C) during quiescent periods in the mission timeline. The EaglePicher Li/CF x -MnO 2 cell design provided a combination of adequate rate capability at moderate capacity over a range of temperatures, making it a suitable choice for operations in the 0 and −40 • C range.
Representative discharge data collected for the baseline Ea-glePicher Li/CF x -MnO 2 cells at various currents at −40 • C, and over a range of temperatures at 50 mA, are provided in Figs. 5 and 6, respectively. Fig. 5 illustrates the effect of temperature increase within the cell during high current discharge. Increasing the discharge current from 50 to 250 mA results in a significant loss of capacity, however increasing discharge current further to 600 mA, 2 A, or 3 A increases delivered capacity at −40 • C. Although the cells were discharged in a convectively cooled chamber, their temperature still rose considerably during discharge, particularly for the 2 A and 3 A discharge conditions. Fig. 6 indicates that, while the cell can support discharge at −40 • C, very little of the CF x portion of the cathode is actually utilized (no second plateau is visible) and the specific energy is approximately half of what is available at 0 • C. Given these performance capabilities over a broad range of operating conditions, this cell chemistry and design was chosen for further development. The other hybrid cell variant (Ultralife) featured a lower specific energy under similar conditions (Table II) in spite of its higher delivered capacity (Figs. 3 and 4). This is primarily due to the mass of the stainless steel can used for the cell design. Use of aluminum cans offers a higher cell level specific energy due to its lower mass vs. stainless steel, but also entails more challenging sealing/welding operations. Several design features were therefore introduced into developmental EaglePicher test cells to optimize performance under the anticipated mission operating conditions, including the use of an aluminum can.    One specific cell design variant featured a lower salt concentration (2R) to reduce solvent viscosity effects at lower temperatures and further improve electrolyte conductivity. A second cell variant featured a higher CF x :MnO 2 ratio along with a lower LiClO 4 salt concentration (3R), to increase the specific energy over a broad temperature range. A third cell design variant (4R) featured this higher CF x :MnO 2 ratio along with a modified solvent blend formulation using FEC as an additive to the standard PC, DME, and THF solvent blend. This modified solvent blend was selected based on earlier developmental studies using three-electrode laboratory cells 7 which suggested that addition of FEC to the electrolyte improves the anode film properties (with a threefold decrease in resistance observed at lower temperatures). These cell design modifications are summarized in Table III, with the delivered specific energy given in Tables II and III. Constant current discharge testing of these modified cells indicated improved specific energy at both 50 mA and 250 mA at 0 • C. In the case of the 4R cell design, the specific energy increased from 437 Wh/kg in the baseline cell to 564 Wh/kg for the advanced development cell at the baseline condition of 50 mA and 0 • C, and increased from 418 Wh/kg to 515 Wh/kg at 250 mA at the same temperature. At −40 • C, however, the same cell design failed to discharge at either rate. The 2R and 3R cell designs, on the other hand, were able to support discharging at both rates at −40 • C. The specific energy of the developmental cells at −40 • C was lower than the baseline version at 50 mA, but higher at 250 mA. The cell discharge voltage vs. specific energy data at 250 mA and 0 • C are given in Fig. 7. The room temperature impedance of cells was evaluated at 1 kHz (Fig. 8). The baseline cells displayed the lowest average 1 kHz impedance of ∼58 m , whereas the 4R variant was found to have the highest average value of >200 m . The EIS responses for three of the EaglePicher Li/CF x -MnO 2 cell variants are shown in Fig. 9, along with a summary of the extracted impedance parameters in Table IV. The "series resistance" refers to the high frequency x-axis intercept (likely dominated by the lead resistance and electrolyte resistance), whereas the "film" resistance refers to the difference between the mid frequency x-axis intercept (inferred from the arc of the semicircle) and the series resistance, and is likely dominated by the film impedance on one of the electrodes (likely the Li anode, as in other Li primary cells). Given the complex nature of the cell chemistry and the limited access to individual cell components, a more sophisticated model is not presented here; detailed discussion can be found in the literature. [8][9][10][11] The 3R cells displayed a significantly higher series resistance relative to the baseline cell design (from 45 to 115 m ), whereas the  "film" resistance remains essentially constant at about 60 m . Comparison of the 4R cells to the baseline cells reveals that both series resistance and "film" resistance are substantially greater in the 4R variant. These data indicate that the new electrolyte in the 4R cells has resulted in a significantly more resistive film on one or both electrodes. Investigation of the discharge profile for the 3R and 4R cells (Fig. 7) reveals this presumed electrode film effect in the 4R cells does not significantly affect the cell voltage under dynamic conditions (during discharge) and is likely removed early in discharge. Although electrochemical impedance spectroscopy (EIS) provides a more complete understanding of the impedance properties of the cell by separating out high and low frequency contributions to the signal, a single 1 kHz point can also be used as an efficient screening tool for large quantities of primary cells. Based on our analysis, the 1 kHz impedance data point correlates with both the "series" and "film" resistances measured by EIS.
The hybrid cell design offers improved performance relative to a pure Li/CF x design over a wider temperature range, resulting from the combination of improvements in materials and electrode properties. 5 As noted earlier, the use of an aluminum can leads to further increases in specific energy, relative to cells using a stainless steel design. The ∼10% higher capacity and minor voltage advantages of the Ultralife hybrid cells were transformed into a nearly 20% specific energy deficit relative to EaglePicher at 50 mA and 0 • C, mainly due to this difference in cell hardware. Increases in the amount of CF x in the cathode (for the developmental cells, relative to baseline designs), combined with the use of the FEC additive, were also found to improve the specific energy of these cells under certain low rate/low temperature conditions. The Li/FeS 2 cells appear to provide the highest specific energy at low temperature (as seen in Table II), although given the sample size and the standard deviation of the data, the performance is comparable to that of Li/SO 2 cells. Recent vendor changes include modifications to the electrolyte, to improve conductivity over a wide temperature range. 12 Electrolytes used in this cell design are typically a mixture of 1,3-dioxolane and 1,2-dimethoxyethane, resulting in formation of a high conductivity and a thin, stable anode film suited to low temperature discharge over a range of rates.

Conclusions
Prior deep space probes into the atmosphere of Jupiter (Galileo Probe) and to Saturn/Titan (Huygens Probe) featured Li/SO 2 cells as the sole power source. Given the emerging mission requirements for a planned Europa lander requiring weeks (rather than hours) of continuous operation, the specific energy of this battery chemistry option is too low to support a viable landed vehicle design. At anticipated operational temperatures centered at ∼0 • C, these cells deliver a specific energy of approximately 280 Wh/kg. Given a mission energy requirement of approximately 25 kWh, this would lead to a full battery mass (with packaging) exceeding 100 kg, which is too large for emerging lander designs. Therefore, despite the flight heritage of this battery chemistry, new cell designs are needed to support the challenging mission power requirements posed by future lander mission concepts.
New cell designs were developed targeting the baseline operating condition of 50 mA and 0 • C, with one of the developmental Li/CF x -MnO 2 cell variants (4R) delivering 564 Wh/kg. This represents a twofold improvement relative to the heritage Li/SO 2 cell chemistry and a nearly 30% improvement relative to the baseline Li/CF x -MnO 2 cell design under the same conditions. Although not specifically designed for lower temperature operation, two available Li/CF x cell designs were evaluated as part of this study, given the need for a very high specific energy. A Rayovac pure Li/CF x D-size cell provided an even higher specific energy (640 Wh/kg) relative to Li/CF x -MnO 2 cells, although a Panasonic C-size Li/CF x cell provided a much lower specific energy (297 Wh/kg). Vendor-provided data indicates the Rayovac cells feature an electrolyte comprising LiBF 4 in PC and DME. 13 No specific cell design data is available for the Panasonic cell, therefore it is difficult to discern the nature of the observed differences in specific energy. Neither the high energy developmental EaglePicher Li/CF x -MnO 2 cell (4R variant) nor the Rayovac Li/CF x cell were functional at −40 • C, which may be required of future surface mission spacecraft designs. Other cell designs operated over the full range of temperatures as noted above, albeit with a lower specific energy. These findings suggest future approaches for cell development, particularly focusing on improved pure Li/CF x cell chemistries (where high specific energy at 0 • C is required) and even Li/FeS 2 (where extended operation at −40 • C is critical).
Several cell options were identified to meet operational requirements over this full temperature range, including both currently available and developmental Li/CF x -MnO 2 cells. In addition, commercially available cells featuring the solid cathode Li/FeS 2 cell chemistry were observed to deliver a high specific energy at −40 • C, comparable to that of the liquid cathode Li/SO 2 system. As the mission operating profiles continue to evolve, cell designs will be further developed and optimized to meet the specific energy and power needs. Future cell performance parameters requiring further investigation include estimation of self-discharge during the very long interplanetary cruise times (5-8 years) and tolerance to radiation, which will be discussed in future publications. The results of this study, however, confirm that developmental Li/CF x -based cell chemistries can exceed the performance of technologies used in prior deep space missions. They have the potential to provide significant benefits to future missions, with respect to supporting much longer mission durations.